Dynamic Event-Triggered Fault-Tolerant Control for Nonlinear Systems by Using Adaptive Dynamic Programming

This article proposes an innovative inputconstrained $H_{\infty}$ fault-tolerant control (FTC) method, which integrates sliding mode control (SMC) techniques with a dynamic event-triggering mechanism. In order to alleviate the effects of time-varying actuator faults, an appropriate SMC method is developed for the system. Meanwhile, a dynamic event-triggered control (DETC) approach is formulated via reinforcement learning (RL) for the equivalent sliding mode dynamics (SMD). To address the computational challenges inherent in solving the Hamilton-Jacobi-Bellman (HJB) equation, a unitary neural network (NN) architecture is implemented to generate a numerical approximation of its solution. This approach facilitates the derivation of both a time-driven worst-case perturbation protocol and a dynamic event-triggered optimal control paradigm. The Lyapunov candidate function is used to analyze the closed-loop systems stability. Additionally, rigorous validation of the methodology's viability and performance superiority through systematically designed simulation.